Guidance on Default assumptions used by the EFSA ...PUBLIC CONSULTATION DRAFT guidance on Default assumptions 3 52 Default Values proposed for use by EFSA Scientific Panels and Committee,

PUBLIC CONSULTATION

DRAFT guidance on Default assumptions

Suggested citation: EFSA Scientific Committee; Guidance on default assumptions used by the EFSA Scientific Panels and Committee, and EFSA Units in the absence of actual measured data. Available online: www.efsa.europa.eu © European Food Safety Authority, 2011 1

ENDORSED FOR PUBLIC CONSULTATION 1

SCIENTIFIC OPINION 2

Guidance on Default assumptions used by the EFSA Scientific Panels and 3 Committee, and EFSA Units in the absence of actual measured data1 4

EFSA Scientific Committee2, 3 5

European Food Safety Authority (EFSA), Parma, Italy 6

ABSTRACT 7

(Max. 300 words, no paragraph breaks. Note that the abstract should end with the copyright) 8

© European Food Safety Authority, 20YY 9

10

KEY WORDS 11 (Max. seven key words are suggested) 12

13

1 On request from EFSA], Question No EFSA-Q-2010-00221, adopted on DD Month YYYY. 2 Scientific Committee members: Boris Antunović, Sue Barlow, Andrew Chesson, Albert Flynn, Anthony Hardy, Klaus-

Dieter Jany, Michael Jeger, Ada Knaap, Harry Kuiper, John Christian Larsen, David Lovell, Birgit Nørrung, Josef Schlatter, Vittorio Silano, Frans Smulders and Philippe Vannier. Correspondence: [email protected]

3 Acknowledgement: The Panel wishes to thank the members of the Working Group on Default Assumptions: Diane Benford, Mikolaj Gralak, Thomas Haertle, Ada Knaap (Chair), John Christian Larsen and Josef Schlatter for the preparatory work on this scientific opinion, and EFSA staff: Davide Arcella, Bernard Bottex, Joanne Gartlon, Frederique Istace, Georges Kass and Istvan Sebestyen for the support provided to this scientific opinion.

PUBLIC CONSULTATIONDRAFT guidance on Default assumptions

2

SUMMARY 14

A number of assumptions and default values are usually applied at the various steps of the risk 15 assessment process. These can be of a methodological nature, or to compensate for the absence of 16 data, in which case the risk assessor may have to refer to default values to be able to perform the 17 assessment. These default values should be scientifically justified and, where possible, be based on 18 existing data. 19

The European Food Safety Authority (EFSA) asked the Scientific Committee to develop a guidance 20 document and to propose harmonised values/procedures to be used as default by the EFSA Scientific 21 Panels, Committee and Units where needed. An internal consultation was organised during the second 22 half of 2009 to review the various default values used within EFSA. The Scientific Committee then 23 identified areas where harmonised default values are needed, and reassessed the extent to which these 24 default values were science-based, making use of the statistics derived from the EFSA Comprehensive 25 European Food Consumption Database and other relevant data sets. Based on this analysis, 26 harmonised default values listed in the table below and default procedures are recommended for the 27 use within EFSA’s Panels and Units in the absence of empirical data. 28 29 The Scientific Committee also considered the use of probabilistic distributions of uncertainty factors 30 (UFs), as an alternative for multiplying various UFs, in establishing health-based guidance values. The 31 Scientific Committee recommends that these probabilistic approaches to combine UFs are further 32 investigated for its potential use in EFSA’s risk assessments. 33 34 The Scientific Committee also recommends the following rules to be applied for rounding figures 35 when establishing health-based guidance values: 36

• The degree of precision for measured values is determined by the precision of the analytical 37 methodology 38

• When reporting derived values, then the degree of precision should take into account the 39 precision of the components used in the derivation. 40

• Rounding figures should be done at the latest stage in the assessment, e.g. when establishing 41 health-based guidance values 42

• Derived values, such as health-based guidance values, will be rounded to a single significant 43 figure if the impact of rounding is less than 10%, and to two significant figures if the impact 44 of rounding to one significant figure exceeds that percentage. 45

46 47 The Scientific Committee underlines that the purpose of this guidance document is harmonisation of 48 default values used by EFSA’s Scientific Panels, Committee and Units, and not standardisation: it is 49 therefore always possible to deviate from the proposed default values, provided that the rationale for 50 such deviation is described. 51

PUBLIC CONSULTATION DRAFT guidance on Default assumptions

3

Default Values proposed for use by EFSA Scientific Panels and Committee, and EFSA Units 52 53

Issue for harmonisation Default value proposed Remark Body weight (Kg) Adults: 70 Infants & children: 12 Total solid food intake No default value Total liquid intake (L) Adults: 2 Children: no default values For children,

See http://www.efsa.europa.eu/en/efsajournal/doc/1459.pdf Converting test compound concentrations in feed (mg/kg), into daily dose (mg/kg b.w. per day)

Rat Chronic studies: 0.05 Subacute studies: 0.12 Subchronic studies: 0.09

Mice Chronic studies: 0.15 Subacute studies: 0.2 Subchronic studies: 0.2

If a subacute or a subchronic study starts at a later age, then the conversion factor for chronic studies should be applied

Converting test compound concentrations in drinking water (mg/l), into daily dose (mg/kg b.w. per day)




UF for inter-species extrapolation

No data on kinetics and/or dynamics available: 10

Variability in toxicokinetics: 4.0 Variability in toxicodynamics: 2.5

UF for intra-human extrapolation


variability in toxicokinetics: 3.16 variability in toxicodynamics: 3.16

UF for Deficiencies in the data available for the assessment

No default UF

Consider the possibility/feasability of getting additional data first. If not feasible, use additional UF (value determined on a case-by-case basis).

UF for for duration of exposure extrapolation

Subchronic to chronic: 2 Subacute to chronic: no default UF

UF to account for the absence of a NOAEL

No default UF Use the BMD approach. If not possible, consider use of the LOAEL with an additional UF (value determined on a case-by-case basis)

UF to account for the severity and nature of the effect

No default UF Genotoxic and carcinogenic compounds: Use the Margin of Exposure approach

Usually not needed. If exceptionally considered necessary, UF value determined on a case-by-case basis.

UF: Uncertainty Factor 54


4

TABLE OF CONTENTS 55

Abstract .................................................................................................................................................... 1 56 Summary .................................................................................................................................................. 2 57 Table of contents ...................................................................................................................................... 4 58 Background as provided by EFSA ........................................................................................................... 5 59 Terms of reference as provided by EFSA ................................................................................................ 5 60 Assessment ............................................................................................................................................... 6 61 1. Introduction ..................................................................................................................................... 6 62 2. Default values for human body weight ............................................................................................ 6 63

2.1. Default body weight value for adults ...................................................................................... 7 64 2.2. Default body weight value for children .................................................................................. 8 65

3. Default values for food intake ......................................................................................................... 9 66 3.1. Daily total solid food intake by humans ................................................................................. 9 67 3.2. Daily total liquid intake by humans ...................................................................................... 10 68

4. Factors for converting chemical compound concentrations in feed or drinking water into daily 69 doses in experimental animal studies ..................................................................................................... 11 70

4.1. Conversion factors for dietary administration of test compounds ........................................ 12 71 4.2. Conversion factors for administration of test compounds in drinking water ........................ 14 72

5. Using animal data for risk assessment – Uncertainty factors ........................................................ 16 73 5.1. Intra/inter-species extrapolation ............................................................................................ 16 74 5.2. Deficiencies in the data available for the assessment. .......................................................... 17 75

5.2.1. Extrapolation for duration of exposure ............................................................................. 18 76 5.2.2. Accounting for the absence of a NOAEL ......................................................................... 19 77

5.3. Severity and nature of the observed effect ............................................................................ 19 78 5.4. Probabilistic approaches and combinations of uncertainty factors ....................................... 20 79

6. Rounding of figures when deriving health-based guidance values ............................................... 22 80 Conclusions and recommendations ........................................................................................................ 23 81 References .............................................................................................................................................. 26 82 Appendix ................................................................................................................................................ 29 83 abbreviations .......................................................................................................................................... 30 84

85


5

BACKGROUND AS PROVIDED BY EFSA 86

In the absence of empirical data, default values are often used to substitute for essential information to 87 perform risk assessments in many of the different areas in the remit of EFSA. Default assumptions 88 may be made at different steps in the risk assessment process, such as in consumer exposure 89 assessment, in converting a feed concentration into a dose in experimental animals or in applying 90 uncertainty factors for extrapolation of animal data to the human situation. In the framework of the 91 transparency activities of the Scientific Committee, a need for harmonisation of the approaches to 92 default assumptions used within EFSA has been identified and the Dietary and Chemical Monitoring 93 (DCM – former DATEX) Unit was asked to prepare, in consultation with the EFSA Panels and Units, 94 a “state of the art” document describing default assumptions presently in use within the remit of 95 EFSA’s activities. 96

The consultation was performed via a questionnaire addressed to all EFSA Panels and Units. The 97 analysis of the responses revealed a considerable degree of similarity for many default assumptions 98 used by the different EFSA Panels and Units in risk assessment. Some default assumptions more 99 specific and of interest for a limited number of Panels and Units were also identified. Based on the 100 current analysis, a fine tuning of similar default assumptions being used by several Panels and Units 101 was recommended. 102

TERMS OF REFERENCE AS PROVIDED BY EFSA 103

Following the suggestion of the Scientific Committee for a self task on the topic of harmonisation of 104 default assumptions used by the different EFSA Scientific Panels and Committee, and EFSA Units, 105 the European Food Safety Authority requests the Scientific Committee to: 106

Develop by end December 2011 a guidance document proposing harmonised values/procedures to be 107 used as default by the EFSA Scientific Panels and Committee, and EFSA Units where needed. 108

Taking into account the outcome of the DCM survey on default assumptions used by EFSA Scientific 109 Panels and Units (areas with consensus, exceptions), the Scientific Committee is requested: 110

• To consider when default assumptions are needed 111

• Where default values are set (e.g. by legal requirements), to discuss whether they are 112 scientifically justified and, if not, to propose some science-based default assumptions. 113

• To compile a table with values/procedures to be used as default within EFSA 114

115

116


6

ASSESSMENT 117

1. Introduction 118

A number of assumptions and default values are usually applied at the various steps of the risk 119 assessment process. These can be of a methodological nature, such as considering the 95th percentile 120 for representing the high consumption when assessing exposure, or to compensate for the absence of 121 data, e.g. when deciding that results obtained from animal experiments can be extrapolated to humans. 122

In cases of insufficient or absence of numerical data, the risk assessor may have to refer to default 123 values to be able to perform the assessment. These default values should be derived on the basis of 124 existing data and be therefore scientifically justified. 125

An internal consultation was organised by the EFSA Dietary and Chemical Monitoring (DCM – 126 former DATEX) Unit during the second half of 2009 to review the various default values used by the 127 EFSA Scientific Panels, Committee and the EFSA Units. The findings highlighted several cases where 128 different default values are used for a same parameter. Recommendation was therefore made to the 129 Scientific Committee to consider whether further harmonisation in the use of default values within 130 EFSA is possible. 131

When reviewing the default values reported by the EFSA Units, the Scientific Committee looked the 132 extent to which they were science-based and could therefore be proposed for the use of EFSA Panels 133 and Units. The Scientific Committee made use of the statistics derived from the EFSA Comprehensive 134 European Food Consumption Database (Comprehensive Database) published on the EFSA website4. 135

It is underlined that the purpose of this guidance document is harmonisation of default values used by 136 EFSA Scientific Panels and Committee, and the EFSA Units, and not standardisation: it is therefore 137 always possible to deviate from the proposed default values, provided that the rationale for such 138 deviation is described. 139

140

2. Default values for human body weight 141

A default value of 60 kg has generally been used by the World Health Organisation (WHO) for 142 calculation of Acceptable Daily Intakes (ADIs) and water quality guidelines (IPCS, 1987; WHO, 143 1994; WHO, 2009), and this value has been adopted in the work of some EFSA Panels. However, the 144 WHO value is intended to apply worldwide, and is not necessarily representative of EU adult 145 populations. The EFSA Pesticides Unit (PRAS - formerly the PRAPeR and PPR units) uses a default 146 value of 60 kg for adult body weight based on the 5th percentile of the distribution of body weights of 147 English adult males (ECETOC, 2001), presumably based on the assumption that use of the 5th 148 percentile is a conservative approach. 149

Assumptions regarding body weight of humans may be required under a number of different 150 circumstances, such as: 151

a) Dietary exposure data are available on a per person basis, and there is a need to convert to a 152 body weight basis in order to compare with an ADI or an Acute Reference Dose (ARfD) (e.g. 153 feed additives, pesticides…) 154

b) Tolerable Daily Intake (TDI) values, expressed on a body weight basis, need to be related to 155 amount of food consumed on a per person basis at different ages, in order to assess safety of 156

4 See http://www.efsa.europa.eu/en/datex/datexfooddb.htm


7

current or proposed maximum levels of a chemical in food (e.g. Specific Migration Limits for 157 food contact materials). 158

c) Non-dietary pesticide exposure estimated per person on the basis of default assumptions is 159 converted to body weight basis in order to compare with an Acceptable Operator Exposure 160 Level (AOEL). 161

d) Dietary exposure resulting in human illness is reported on a per person basis and needs to be 162 converted to a body weight basis in order to establish health-based guidance values (e.g. 163 ARfDs for some marine biotoxins). 164

165

In situations (a), (b) and (c) above, the implication of underestimating body weight is conservative, i.e. 166 it will result in overestimation of exposure, and hence of risk. In contrast, for situation (d) 167 underestimating body weight is not conservative, since it would result in a higher health-based 168 guidance value. Therefore it cannot be assumed that application of a low default value for body weight 169 is always a worst case scenario. The Scientific Committee decided that a realistic estimate of typical 170 body weight for the EU population would support a more proportionate approach to risk assessment. 171 Typical body weight is best represented by the median of the distribution. Variation between 172 individuals is allowed for by the application of uncertainty factors in setting health-based guidance 173 values. 174

175

2.1. Default body weight value for adults 176

The Scientific Committee used the Comprehensive Database, which was built from existing national 177 information on food consumption at a detailed level. Competent organisations in the European 178 Union’s Member States provided EFSA with data from those most recent national dietary surveys in 179 their country, at the level of consumption by the individual consumer. This included food consumption 180 data concerning infants (from 0 to less than 3 months, 3 to less than 6 months, 6 to less than 12 months 181 old), toddlers (from 1 to less than 3 years old), children (from 3 to less than 10 years old), adolescents 182 (from 10 to less than 14 years, 14 to less than 18 years old), adults (from 18 to less than 65 years old), 183 elderly (from 65 to less than 75 years old and very elderly (75 years old and over), for a total of 32 184 different dietary surveys carried out in 22 different Member States. 185

Table 1 summarises the body weight data for adult subjects in the EFSA Comprehensive Database. All 186 medians are higher than the currently applied default value of 60 kg, with the lowest median body 187 weight for females aged 18-64 years (66kg). The median body weights for older females, and for all 188 age groups of men are larger. Taking into account also that body weights are tending to increase, the 189 Scientific Committee concluded that a default value of 70 kg is a closer approximation to the typical 190 body weight of the EU adult population than the currently applied default value of 60 kg. The 191 Scientific Committee therefore recommends that 70 kg be used by EFSA when there is a need to apply 192 a default body weight value for adults. 193

194

195


8

Table 1: Body weight (kg) statistics for adult subjects in all surveys of the EFSA Comprehensive 196 database 197

Age (years) Gender N Mean StdDev Median P5 P95

% ≤ 70kg

% > 70kg

18-64 ♀ 22507 67.2 12.5 66.0 50.0 91.0 71.0 29.0 18-64 ♂ 18699 82.1 13.6 82.0 62.6 105.0 18.5 81.5 18-64 ♀+♂ 41206 74.4 15.0 73.0 52.6 100.0 52.9 47.1 65-75 ♀ 2343 70.6 12.1 71.0 53.0 92.0 49.0 51.0 65-75 ♂ 2066 82.2 11.5 82.7 65.0 103.0 14.6 85.4 65-75 ♀+♂ 4409 76.0 13.2 75.0 55.0 98.9 67.1 32.9 ≥75 ♀ 1230 66.8 12.6 67.0 50.0 86.0 62.4 37.6 ≥75 ♂ 1030 78.1 11.8 79.0 60.0 96.0 30.5 69.5 ≥75 ♀+♂ 2260 71.4 13.6 70.4 50.0 92.0 52.2 47.8

N: number of individuals in the database 198 StdDev: standard deviation 199 Pxx: xxth percentile 200 201

2.2. Default body weight value for children 202

Infants and young children have higher food intake than older children or adults expressed on a body 203 weight basis due to their higher energy requirement during rapid growth (see Table 2). The highest 204 total food intake seen in table 2 is for infants aged 3-6 months. However at this age the diets are 205 specialised, comprising mainly breast or formula milk, with possible gradual introduction of small 206 amounts of a limited number of foods. The age group with the next highest food intake is the 207 “toddlers” (aged 1-3 years), and at this age the food consumed is more similar to that of other age 208 groups. Therefore the 1-3 year age-group should be included in exposure assessments as they are 209 likely to have the highest dietary exposure expressed on a body weight basis. 210

Table 2: Total food (solid + liquid) consumption (g/kg b.w. per day) statistics for infants and 211 children in all surveys of the EFSA Comprehensive database 212

Age range Total food consumption

(g/kg b.w. per day) Number of individuals in

database Infants [0-3 months[ 103.0 205 Infants [3-6 months[ 132.4 231 Infants [6-12 months[ 106.9 441 Toddlers [1-3 years[ 114.4 1679 Other children [3-10 years[ 74.1 8833 Adolescents [10 - 14 years[ 43.5 3291 Adolescents [14 - 18 years[ 37.0 3935 Adults [18-65 years[ 37.5 41206 Elderly [65-75 years[ 36.7 4409 Very elderly (≥75 years) 32.7 2260 213

Table 3 summarises the body weight data for infants and children in the EFSA Comprehensive 214 Database. The body weights of girls and boys are very similar up to the age of 14 years and are 215 combined for statistical purposes in the table. The Scientific Committee recommends that, since the 216 dietary exposure of infants aged 1-3 years is likely to be higher than that of older children, a default 217 body weight of 12 kg based on the median body weight of 1-3 year olds could be used in a 218


9

conservative approach for all infants and children. However if deviation from the default value is 219 required for assessments for specific age groups, the median values identified in table 3 should be 220 applied. 221

The Scientific Committee noted that PRAS uses different age categories in their assessments, based on 222 ECETOC Report No 79 (2001), and advised that their approach should be aligned with the age 223 categories in the EFSA Database. 224

Table 3: Body weight (kg) statistics for infants and children in all surveys of the EFSA 225 Comprehensive database 226

Age (years) Gender N Mean StdDev Median P5 P95 Infants (0-3 months) ♀+♂ 205 4.8 1.4 4.8 3.2 6.4 Infants (3-6 months) ♀+♂ 231 6.7 1.0 6.7 5.1 8.5 Infants (6-12 months) ♀+♂ 441 8.8 1.2 8.7 7.0 11.0 Toddlers (1-3 years) ♀+♂ 1679 11.9 2.2 11.6 8.7 15.9 Other children (3-10 years) ♀+♂ 8833 23.0 7.1 21.6 14.0 36.3 Adolescents (10-14 years) ♀+♂ 3291 43.2 10.5 42.0 29.0 62.0 Adolescents (14-18 years) ♀ 2048 57.4 9.3 56.0 45.0 76.0 Adolescents (14-18 years) ♂ 1887 65.3 11.9 65.0 47.0 87.0 Adolescents (14-18 years) ♀+♂ 3935 61.2 11.3 60.0 45.0 83.0

Abbreviations: see table 1’s footnotes 227 228

Conclusions: 229

• A body weight of 70 kg should be used as default for European adults. 230

• For dietary exposure assessment, a body weight of 12 kg should be used as default for 231 European infants and children. If deviation from the default value is required for the 232 assessment of specific age groups, the median values identified in table 3 should be used. 233

234

3. Default values for food intake 235

3.1. Daily total solid food intake by humans 236

Relevant empirical data for EFSA Panel risk assessments seem in most cases to be available from food 237 consumption databases, e.g. the EFSA Comprehensive Database, and being already commonly used, 238 they negate the need for default values. There are some examples of EFSA Panels, such as the Panel 239 on Food Additives and Nutrient Sources Added to Food (ANS), which has been using until recently 240 the concept of total solid food (i.e. Budget Method) as an initial step in the exposure assessment. 241 However the ANS Panel will now base its exposure estimates on data from the EFSA Comprehensive 242 Database coupled with maximum permitted use levels and, if available, maximum reported use levels 243 or analytically determined use levels. 244

Rather than considering daily total solid food intake, intake values for specific food categories (i.e. 245 those containing the substance in question) tend to be more relevant for the risk assessments 246 performed by some EFSA Panels (e.g. the Panel on Additives and Products or Substances used in 247 Animal Feed - FEEDAP) and in some cases the default values used are defined in EU legislation. For 248


10

example, Commission Regulation (EC) No 429/2008 of 25 April 20085 regarding feed additive 249 applications provides theoretical daily human consumption figures for tissues and products and 250 potential consumer exposure to the animal feed additives and/or metabolite(s) is calculated based on 251 these consumption figures. However, this Regulation also states that ‘in certain situations (e.g. some 252 nutritional and sensory additives or additives intended for minor species) it may be appropriate to 253 subsequently refine the human exposure assessment using more realistic consumption figures, but still 254 keeping the most conservative approach. Where this is possible this shall be based on Community 255 data.’ Consequently, the FEEDAP Panel is reviewing the theoretical consumption figures to be used in 256 exposure assessments and any modifications will be included in subsequent FEEDAP guidance 257 documents for use by applicants. 258

Some EFSA Panels refer to guidance documents containing relevant default values specific for their 259 risk assessments. For example, the Panel on Food Contact Materials, Enzymes, Flavourings and 260 Processing Aids (CEF) use the Guidelines of the Scientific Committee on Food for the presentation of 261 an application for safety assessment of a substance to be used in food contact materials prior to its 262 authorisation (EC, 2001) which “… maintains the assumption that a person may consume daily up to 1 263 kg of food in contact with the relevant food contact material.” The CEF Panel is currently reviewing 264 this guideline and considering other approaches for this consumption estimate. Guidance documents 265 for risk assessments used by EFSA Panels are available on the EFSA website 266 (http://www.efsa.europa.eu/). 267

268

Conclusion: 269

As the various EFSA Panels have different considerations and methodologies for their risk 270 assessments, it was concluded that a single default value for daily total solid food intake for adults for 271 harmonised use across the EFSA Panels was neither needed nor justifiable. The SC recommends that 272 EFSA Panels regularly consult the EFSA Comprehensive Database in order to check the relevance of 273 default values used and to be fully aware of their impact on exposure estimates. The Scientific 274 Committee considers that default values used in EFSA Guidance documents, which are considered to 275 more accurately reflect consumption/exposure, should also be considered for inclusion in the 276 respective legislation. The Scientific Committee did not consider possible default values for daily total 277 solid food intake for children, as the EFSA Panels and Units did not report using such default values. 278

279

3.2. Daily total liquid intake by humans 280

A number of EFSA Panels use default values for total liquid intake in the absence of empirical data. In 281 general, the default value of 2 L total liquid intake per day for adults is used (e.g. the EFSA Panel on 282 Plant Protection Products and their Residues (PPR) follows EC guidance6 which stipulates 2 L water 283 consumption is used for calculating human exposure to substances of unknown structure for applying 284 a Threshold of Toxicological Concern (TTC)). The EFSA Comprehensive consumption database 285 contains data from different EU member states on daily total liquid intake, but due to different 286 methodologies used to collect and report total liquid intake data in the various studies, these data sets 287 cannot be aggregated in order to determine the relevance of a default value of 2 L for total liquid 288 intake for European adults. 289 290 5 Commission Regulation (EC) No 429/2008 of 25 April 2008 on detailed rules for the implementation of Regulation (EC) No 1831/2003 of the European Parliament and of the Council as regards the preparation and the presentation of applications and the assessment and the authorisation of feed additives 6 Guidance Document on the Assessment of the Relevance of Metabolites in Groundwater of Substances Regulated under Council Directive 91/414/EEC


11

In a recent opinion on dietary reference values for water, the Panel on Dietetic Products, Nutrition and 291 Allergies (NDA) defined adequate intakes of total water for adults (≥14 years old) using a 292 combination of observed intakes in population groups, desirable osmolarity values of urine and 293 desirable water volumes per energy unit consumed (EFSA NDA Panel, 2010). The total water intake 294 values include water from drinking water, beverages of all types, and from food moisture and only 295 apply to conditions of moderate environmental temperature and moderate physical activity levels. 296 Based on median potential renal solute loads calculated from dietary surveys from 4 European 297 countries, the NDA Panel calculated that to achieve a urine osmolarity of 500 milli-osmoles 298 (mosm)/L, male adults would need urine volumes of 2 L and female adults would need urine volumes 299 of 1.6 L. The NDA panel calculated that this urine volume should be attained from a total water intake 300 (i.e. drinking water, beverages of all kind, and from food moisture) of 2.5 L for males and 2 L for 301 females. The NDA Panel recommended the same adequate intakes of total water for elderly adults as 302 for adults, as despite reduced energy requirements the water requirement per energy unit from the diet 303 becomes greater due to a decreased capacity for renal concentrating (EFSA, 2010). The NDA opinion 304 (EFSA, 2010) also gives adequate intake levels of total water for children according to their age. 305

The calculated values in the NDA opinion are in line with the default value of 2 L used in WHO risk 306 assessment guidance documents such as Environmental Health Criteria - EHC 240 on ‘Principles and 307 methods for the risk assessment of chemicals in food’ (IPCS, 2009). EHC 240 refers to the WHO 308 Guidelines on drinking water (WHO, 2008), in giving the default value of 2 L for total daily drinking 309 water consumption for adults (60 kg bw). Earlier WHO EHC publications (IPCS, 1994, 1999) also 310 reported that the WHO used 2 L for total daily drinking water consumption for adults in calculating 311 water quality guidelines, but also referred to the International Commission on Radiological Protection 312 (ICRP) Report ‘Reference Man’ (ICRP, 1974), with EHC 170 stating that the ICRP intake volumes 313 should be used for exposure estimates. In this ICRP Report (1974), it is stated that daily fluid intake 314 values (i.e. milk, tap water, other beverages etc. and excluding food) were derived by considering that 315 1 ml of water is required for each kilocalorie of energy expended (ICRP, 1974). The total daily fluid 316 intakes reported for the ICRP Reference Man are 1950 ml for males (70 kg bw) and 1400 ml for 317 females (58 kg bw), but it was highlighted that actual values for reference man may range from 1000 318 to 2400 ml/day at moderate temperatures (ICRP, 1974). Based on these values, EHC 170 and EHC 319 210 report a reference value of 1900 ml/day daily fluid intake (i.e. milk, tap water, other beverages) 320 for adults (bw not stated). 321 322 Conclusion: 323

Considering the inherent variation in water intakes and in view of the similar values for total liquid 324 intakes reported for adult males and adult females in the NDA opinion (EFSA, 2010) and the ICRP 325 (1974) report, the Scientific Committee recommends using 2 L as a conservative default value for 326 daily total liquid intake in adults to be used in risk assessments. 327

328

4. Factors for converting chemical compound concentrations in feed or drinking water 329 into daily doses in experimental animal studies 330

In dietary risk assessment, health-based guidance values are usually established based on a range of 331 experimental animal studies where the compound of interest is orally applied. This is either done by 332 applying the compound by gavage, via drinking water or, more commonly, via feed. When using 333 gavage application of the compound, the animals are dosed individually and the exact doses are 334 known. However, in the latter two cases, accurate doses can be calculated only when both body weight 335 (bw) and feed or water consumption are reported. Where accurate doses cannot be calculated because 336 of the lack of measured body weights and food or water consumption, according to WHO (2009) 337 approximate doses can be estimated using dose conversion factors for feeding studies. To convert a 338 feed concentration into a dose, for rats and mice a factor of 0.05 and 0.15, respectively, is proposed by 339


12

WHO (e.g. 1 mg/kg in feed is equivalent to a dose of 0.05 and 0.15 mg/kg bw per day in rats and 340 mice, respectively). In addition, for young rats, WHO proposes a conversion factor of 0.1, while for 341 mice, no such distinction was made. These conversion factors were taken unchanged from WHO EHC 342 70 (IPCS, 1987), originating from Lehman (1954). No recommendations for conversion factors for 343 toxicity studies where the compound is administered in the drinking water are available. 344

The most frequent situation where such a conversion is needed for assessing toxicological feeding 345 studies is for studies in rats and mice. Although feeding studies with dogs are also frequently reported 346 (e.g. for pesticide risk assessments), such conversion factors would not be needed as in these studies 347 the dogs are fed individually and the actual doses are usually reported. For this reason the validity of 348 the WHO default values for converting feed concentrations into a dose for rats and mice are explored. 349

350

4.1. Conversion factors for dietary administration of test compounds 351

To see whether the conversion factors proposed by WHO for rats and mice are supported by more 352 recent data, factors for converting chemical compound concentrations in feed (e.g. mg/kg) into doses 353 (e.g. mg/kg b.w. per day) in animal studies were calculated from long-term rat and mouse feeding 354 studies, using control group mean weekly food consumption data and control group mean body weight 355 data from 37 chronic rat and 38 chronic mouse studies. The reason for using only control group data is 356 to avoid possible influence of compound administration on the feed consumption or potential effects 357 on bw. Original data from dossiers submitted to the Swiss Federal Office of Public Health over the last 358 30 years on plant protection products (PPP) were used. These original studies formed also the basis for 359 international evaluations establishing ADIs for these PPP (i.e. EFSA and the Joint FAO/WHO 360 Meeting on Pesticide Residues (JMPR)). All studies reported feed-consumption and body weight data 361 on a weekly basis at least for the first 13 weeks of the study and thereafter every 4 weeks. The start of 362 the studies (i.e. beginning of the treatment of the animals) was usually between weeks 5 and 8 of age 363 and lasted for 104 weeks in rats but mouse studies were often terminated after 80 weeks. 364 Consequently, data were available from 5 weeks onward. In the 37 chronic rat studies the following 365 strains were used: Wistar: 18 studies, Sprague-Dawley: 9 studies, Fischer 344: 8 studies, and CD BR: 366 2 studies. In the 38 chronic mouse studies the following strains were used: CD-1: 28 studies, B6C3F1: 367 2 studies, C57BL: 7 studies, and NMRI: 1 study. 368

First, for each study, weekly conversion factors were calculated by dividing the listed weekly feed 369 consumption by the respective weekly bw. Then, the means and their respective standard deviations of 370 the calculated weekly conversion factors over all studies were calculated. The graphical plot of all 371 weekly means and the respective standard deviations is shown in Figure 1. 372

As can be seen from Figure 1, a rapid drop of the conversion factor occurs over the first few weeks, 373 especially in rats, and it levels off around week 20. Conversion factors calculated for week 5 were 0.15 374 and 0.14 in male and female rats and 0.2 and 0.23 in male and female mice, respectively. In general, 375 the conversion factors for males are lower than those for females and the standard deviations in the 376 mouse data were larger than those from rat data. The increase in the standard deviations seen after 377 week 80 in the mouse studies is due to the rapidly decreasing number of studies lasting over the full 378 period of 104 weeks. A similar pattern was observed already by Luijckx et al. (1994) based on eight 2-379 year carcinogenicity (feeding) studies in Fischer 344 rats performed by the U.S. National Toxicology 380 Program (NTP) by using a slightly different approach. 381


13

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

105

Age of animals (weeks)

conv

ersi

on fa

ctor

male ratfemale ratmale mousefemale mouse

382

Vertical bars indicate the standard deviation of the weekly mean. Note that the start of the treatment of the animals was 383 usually between weeks 5 and 8 of age. Calculated with Microsoft® Office Excel 2003 384

Figure 1: Graphical plot of weekly means of factors for converting chemical compound 385 concentrations in feed into daily doses for male and female rats and mice, respectively. 386

387

From these data, the mean of all weekly means were calculated, representing average conversion 388 factors for chronic (i.e. 104 week) studies. Furthermore, the means for week 5 to 9 of age (4 weeks) 389 and 5 to 17 (13 weeks) were also calculated, representing average conversion factors for subacute (i.e. 390 4 week) or subchronic (i.e. 13 week) studies. The resulting figures are listed for rats and mice in Table 391 4. 392

Table 4: Mean factors for converting concentrations of substances in feed into a daily dose for rats 393 and mice for , subacute, subchronic and chronic study duration. 394

Study type (statistics) Male rat Female rat Male mice Female mice Subacute 0.118 0.117 0.188 0.224 Subchronic 0.081 0.091 0.169 0.215 Chronic 0.045 0.058 0.130 0.167

395

For rats, there are no large differences between males and females in the mean conversion factors for 396 feed concentrations into a dose. For mice, these differences between males and females are seemingly 397 larger but the standard deviations are also larger. Therefore it is considered that no sex specific 398 conversion factors are needed and only one single value for each species is selected. The mean values 399


14

for both sexes are 0.052 and 0.149 for rats and mice, respectively, which are close to the respective 400 values proposed by WHO. 401

402

Conclusions: 403

Based on this analysis it is concluded that the conversion factors proposed by WHO of 0.05 and 0.15 404 for rats and mice respectively, are supported for chronic feeding studies by the data considered. The 405 Scientific Committee recommends that within EFSA these conversion factors are used as defaults to 406 calculate average doses in mg/kg bw per day from feed concentrations in mg/kg feed in the absence of 407 measured actual data. However, it should be noted that the initial dose administered to rats during the 408 first week is 3 times higher than the resulting calculated default dose. The respective ratio for mice is 409 1.5. 410

The Scientific Committee also recommends the following conversion factors to be used as defaults for 411 shorter study durations: for subacute studies 0.12 and 0.2 for rats and mice, respectively, and for 412 subchronic studies 0.09 and 0.2 for rats and mice, respectively. 413

As mentioned earlier, experiments start usually at week 5 to 7. If a subacute or a subchronic study 414 starts at a later age, e.g. after week 25, then the conversion factor for chronic studies should be 415 applied. 416

417

4.2. Conversion factors for administration of test compounds in drinking water 418

As mentioned before, so far there are no recommendations for conversion factors for toxicity studies 419 where the compound is administered in the drinking water. To see whether default conversion factors 420 for such studies can be derived from existing studies, the dossiers of long-term rat and mouse feeding 421 studies used to derive conversion factors for feed were checked for the availability also of weekly 422 drinking water consumption data. Surprisingly, no such data were found with the required level of 423 detail. Therefore, published NTP long-term study reports were searched for studies where the 424 compound was administered in the drinking water. Out of the 573 NTP studies published by 425 December 2010, 20 studies were identified in which the compound was administered in drinking 426 water. However, only in 8 studies out of these 20 studies, drinking water consumption and body 427 weights were reported on a weekly basis for the first 13 weeks of the study and thereafter every 4 428 weeks. The remaining 12 studies did not report the required detailed drinking water consumption and 429 body weights during the first weeks. As the body weight gain during the first weeks of a chronic study 430 is largest and, therefore, any conversion factor is anticipated to change most during this phase, these 431 studies were not considered suitable for the purpose of deriving conversion factors for different study 432 phases. The 10 most recent NTP long-term study reports with dietary compound administration (2004-433 2010) were also searched for detailed data on drinking water consumption. As with the PPP dossiers, 434 no such detailed data were identified. Equally, in the 5 most recent NTP gavage studies, in 2 NTP 435 studies with dermal application and in one inhalation study, no detailed drinking water consumption 436 data were identified. 437

All animals were 5 to 7 weeks old at the start of the study. In 7 of the selected studies, both 438 Fischer344/N rats and B6C3F1 mice were used. In one study, only B6C3F1 mice were tested and in 439 another study, a second rat strain (Wistar) was also tested besides the Fischer344/N rats. 440

Weekly factors for converting drinking water concentrations into daily doses in these 8 studies in rats 441 and mice were calculated, using control group mean weekly drinking water consumption data and 442 control group mean body weight data. The reason for using only control group data is again to avoid 443


15

possible influence of compound administration on the drinking water consumption or potential effects 444 on body weight. As before with the feeding studies, the mean and the respective standard deviations of 445 the calculated weekly conversion factors over all studies were calculated. The graphical plot of all 446 weekly means and the respective standard deviations is shown in Figure 2. 447

drinking water consumption / bw

0

0.05

0.1

0.15

0.2

0.25

0.3

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

study duration (weeks)

conv

ersi

on fa

ctor

male ratfemale rat

drinking water consumption / bw

0

0.05

0.1

0.15

0.2

0.25

0.3

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

study duration (weeks)

conv

ersi

on fa

ctor

male micefemale mice

448 Vertical bars indicate the standard deviation of the weekly means. Note that the x-axis relate to study duration in weeks and 449 not to the actual age of the animals. Animals were 5 to 7 weeks old at the start of the study . Calculated using Microsoft® 450 Office Excel 2003 451

Figure 2: Graphical plot of weekly means of factors for converting chemical compound 452 concentrations in drinking water into daily doses for male and female rats (left plot) and mice (right 453 plot). 454

455

As can be seen from Figure 2, again a rapid drop of the conversion factor occurs over the first few 456 weeks, especially in rats, and it levels off around week 20. Conversion factors calculated for the first 457 week of treatment (i.e. at an average age of the animals of 6 weeks) were 0.16 and 0.15 in male and 458 female rats and 0.19 and 0.2 in male and female mice, respectively. The conversion factors for male 459 and female rats are similar. A similar pattern was also observed by Luijckx et al. (1994), based on four 460 2-year NTP carcinogenicity studies in F344 rats, where the compound was administered in the 461 drinking water by using a slightly different approach. They observed a twofold increase in drinking 462 water consumption towards the end of the male rat studies, which was attributed to the high incidence 463 of nephropathy in the aging male F344 rat. 464

In mice, the conversion factors for males tended to be lower than those for females in the first weeks 465 but were higher after about 25 weeks. 466

From these data, the mean of all weekly means were calculated, representing average conversion 467 factors for chronic (i.e. 104 week) studies. Furthermore, the means for subacute or subchronic (i.e. the 468 first 4 or 13 weeks, respectively) were calculated. The resulting figures are tabulated for rats and mice 469 in Table 5. 470

Table 5: Mean factors for converting concentrations of substances in drinking water into a daily 471 dose for rats and mice for subacute, subchronic and chronic study duration. 472

Study type (statistics) Male rat Female rat Male mice Female mice Subacute 0.125 0.121 0.174 0.191 Subchronic 0.089 0.093 0.144 0.164 Chronic 0.052 0.057 0.103 0.083

473


16

As seen for feed, for rats there are no large differences between males and females in the mean factors 474 for converting concentrations of substances in drinking water into a daily dose. For mice, these 475 differences between males and females are again seemingly larger but the standard deviations are also 476 larger. Therefore it is considered that no sex specific conversion factors are needed and only one 477 single value for each species is selected and the mean values for both sexes used. 478

479

Conclusions: 480

Surprisingly few studies were found where weekly drinking water consumption data were given. 481 Conversion factors for chronic rat and mice studies of 0.05 and 0.09, respectively, could be derived 482 from the data set considered. The Scientific Committee recommends that within EFSA these 483 conversion factors are used as defaults to calculate doses in mg/kg bw per day from concentrations in 484 drinking water in mg/l in the absence of measured actual data. However, as seen with compound 485 concentration conversion factors for feed, it should be noted that the initial dose administered to rats 486 during the first week is 3 times higher than the resulting calculated default dose. The respective ratio 487 for mice can be up to 2.5. 488

The Scientific Committee also recommends the following conversion factors to be used as defaults for 489 shorter study durations: for subacute studies 0.12 and 0.18 for rats and mice, respectively, and for 490 subchronic studies 0.09 and 0.15 for rats and mice, respectively. 491

Again, experiments start usually at week 5 to 7. If a subacute or a subchronic study starts at a later age, 492 e.g. after week 25, then the conversion factor for chronic studies should be applied. 493

494

5. Using animal data for risk assessment – Uncertainty factors 495

Uncertainty factors (UFs) (also called assessment factors, safety factors, adjustment factors or 496 extrapolation factors) are used to derive health-based guidance values (HBGV) by extrapolating from 497 experimental animal data to humans (IPCS, 2009). 498

UFs are intended to cover the uncertainty and variability arising from the inter-species differences, 499 intra-species differences, extrapolation from sub-chronic or subacute to chronic exposure, absence of a 500 No-Observed-Adverse-Effect-Level (NOAEL), quality of the toxicological database, and severity of 501 the effect. Additionally, some considerations will be given about probabilistic approaches and 502 combination of UFs. 503

504

5.1. Intra/inter-species extrapolation 505

The default UF of 100 was introduced in 1954 by Lehman and Fitzhugh, and adopted in 1958 by the 506 Joint FAO/WHO Expert Committee on Food Additives (JECFA). 507

This factor was later interpreted as reflecting extrapolation from experimental animal to human (factor 508 10 for inter-species variability) and extrapolation from an average human NOAEL to a sensitive 509 human NOAEL (factor 10 for human or intra-species variability). 510

A further division of these inter- and intra-species factors into subfactors based on specific quantitative 511 information on toxicokinetics and toxicodynamics has been proposed by WHO/IPCS (2005). This 512 division permits the use of specific data on a chemical to derive chemical-specific adjustment factors. 513


17

(CSAF). Compound specific data for one particular aspect of uncertainty should be used to replace the 514 relevant part of the overall default UF (see table 6). 515

Table 6: Values for default UFs that can be replaced by CSAFs (WHO/IPCS, 2005) 516

Source of uncertainty Default subfactor Toxicokinetic Toxicodynamic Combined Inter-species variation 4.0 2.5 10 Human interindividual variation 3.16 3.16 10 517

EFSA uses body weight as a scale for inter-species extrapolation. Other EU institutions (e.g. European 518 Chemicals Agency (ECHA)) have proposed use of allometric scaling based on caloric demand 519 (metabolic body weight BW0.75). The underlying principle is that due to the faster metabolic rate of 520 small animals, humans would less effectively detoxify and/or excrete xenobiotics than laboratory 521 animals. However, many chemicals of concern rely upon specific enzymes or transporters for their 522 toxicity or elimination that do not scale allometrically. Such alternative scaling has not been used so 523 far within EFSA. As no general consensus has been reached yet on alternative scaling, no 524 harmonisation of this approach can be recommended in this opinion. 525

526

Conclusions: 527

If relevant chemical-specific data on kinetics and/or dynamics are available, the default subfactors 528 listed in table 6 should be considered. In the absence of such data, the Scientific Committee recomends 529 using the overall default UF of 100 (10x10). 530

531

5.2. Deficiencies in the data available for the assessment. 532

Significant data deficiencies may warrant an additional factor due to high level of uncertainty. To take 533 into account the quality of the available database, a transparent expert judgement is important on a 534 case-by-case basis. When the standard data package for a regulated chemical is incomplete (e.g. when 535 endpoints which might prove critical were not measured), this might result in a higher critical NOAEL 536 than would be provided by a more complete data package. Therefore, an additional UF may be 537 required (COT, 2007). 538

In the Guidelines for Drinking Water Quality (WHO, 2008), the application of an additional UF 539 between 1 and 10 is suggested depending on the adequacy of databases. 540

According to EHC 240, the quality of the total database may affect the choice of UF. Significant data 541 deficiencies may warrant an increased factor due to increased uncertainty. No default values are 542 proposed but rather a case-by-case approach pending on the nature of the deficiencies. Alternatively, 543 when the data were not sufficient to propose a HBGV, JECFA has calculated the ratio between an 544 amount of the substance producing a small but measurable effect in laboratory animals or humans and 545 the estimated human dietary exposure, in order to characterize the risks associated with certain 546 contaminants in food. 547

In the case of smoke flavours, because they are complex mixtures of variable and incompletely 548 characterised composition, and in view of the limited toxicological data, the EFSA CEF Panel 549 considered it inappropriate to establish an ADI but calculated a margin of safety based on the NOAEL 550 in a 90-day study (EFSA CEF Panel, 2010). The margin of safety was defined as the ratio between the 551


18

NOAEL of the critical effect in the animal study on the smoke flavouring and the anticipated dietary 552 exposure of consumers to that smoke flavouring. An additional UF relating to the quality of the 553 toxicological database on which the evaluation is based, can be considered for the interpretation of the 554 margin of safety. In those cases, where the overall evaluation of the genotoxicity studies did not raise 555 cause for concern in vivo and where the 90-day studies were of adequate quality by current standards, 556 the CEF Panel considered that, normally, an extra UF of 3-fold in addition to the default UF of 100, 557 should be sufficient to cover the limited duration and statistical power of the pivotal study. Whether a 558 specific margin of safety for a particular smoke flavouring is sufficient is highly dependent on the 559 specific situation (e.g. composition, variability and stability, quality of the toxicological data) and 560 default guidance cannot be given. 561

562

5.2.1. Extrapolation for duration of exposure 563 Different approaches followed by international organizations are described in the literature when 564 considering the extrapolation for duration of exposure (see Table 7). 565

Table 7: Existing UFs used in the extrapolation for the duration of exposure (Falk-Filipsson, 2007) 566

Organization Applicability Extrapolation Value of UF US EPA (2002) Not .defined. Subchronic to chronic 10 EU (EC, 1996) Industrial chemicals Case-by-case based on expert judgement of scientific

information EMEA/ICH (1997) Residual solvents in pharmaceuticals 6-month rodents to chronic 2

3-month rodents to chronic 5 < 3-month rodents to chronic 10

EC (1967) Classification and labelling of chemicals for health effects

Subacute to subchronic 3

ECHA, REACH guidance doc (2010)

REACH chemicals Subacute to subchronic 3 Subchronic to chronic 2 Subacute to chronic 6

Subacute: 28-day, subchronic: 90-day, chronic: 1.5 – 2 years. 567 568

There have been a number of analyses of data where comparison has been made between NOAELs 569 and/or Lowest-Observed-Adverse-Effect-Levels (LOAELs) from sub-chronic and chronic feeding 570 studies, and ratios between the two developed. Diverging results were obtained from these analyses. 571 However studies used for these analyses have substantial differences in their design, influencing the 572 outcome of the analyses (IGHRC, 2003). 573

Recently, Zarn et al. (2010) have suggested taking into account the dose decrement related to 574 decreased food intake during chronic feeding studies, when compared to subchronic feeding studies in 575 rats with plant protection products. They concluded that a chronic rat NOAEL can be accurately 576 predicted by dividing the NOAEL from rat subchronic studies by the dose decrement factor of 1.7 577 between the subchronic and chronic period. 578

579

The EFSA Scientific Committee is not in a position to propose default values to extrapolate from 580 subacute to chronic duration. On the basis of a 28-day subacute study, where few parameters are 581 usually investigated (limited study design), an extrapolation for chronic duration should be considered 582 on a case-by-case basis. Therefore it is not recommended to multiply the factors in table 7 (subacute to 583 chronic (6) = subacute to subchronic (3) x subchronic to chronic (2)). However, taking into 584 consideration that the investigations are more extensive in a 90-day study, the extrapolation from 585 subchronic to chronic duration can be performed as proposed by ECHA (UF of 2), supported by Zarn 586 et al. (2010). 587


19

588

5.2.2. Accounting for the absence of a NOAEL 589 In the case where a NOAEL cannot be identified from the critical toxicological study, the evaluation 590 might have to rely on the LOAEL. Several organisations recommend in their guidelines, e.g. 591 Guidelines for drinking water Quality (WHO, 2008), the application of an additional UF of up to 10 to 592 the LOAEL to derive a health-based guidance value. 593

At the same time, as mentioned in EHC 240, the consideration of the shape of the dose-response curve 594 may trigger the need for an additional UF if the curve is very steep, particularly when the NOAEL is 595 close to the LOAEL. In its guidance (2010), ECHA also suggests to take into account the dose spacing 596 in the experiment, the shape and slope of the dose-response curve, and the extent and severity of the 597 effect observed at the LOAEL in order to determine the size of this additional UF. 598

As recommended in a previous opinion of the Scientific Committee, it is preferable to use the 599 benchmark dose (BMD) approach (EFSA, 2009) instead of the NOAEL/LOAEL. An advantage is 600 that, even in the absence of a NOAEL, the BMD approach can still be applied without the need for an 601 additional UF. The use of the BMD approach is also advocated by ECHA in its guidance on 602 Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) (ECHA, 2010). 603

604

Conclusions: 605

In case of deficiencies in the data for applications, the EFSA Scientific Committee recommends that, 606 rather than applying an additional UF, the possibility/feasibility of getting additional data improving 607 the quality of the dataset is first considered. Therefore, the lack of data should not directly imply the 608 use of an extra UF. 609

When additional data cannot be obtained or requested, the use of an additional UF to take account of 610 the deficiency of a database should be considered on a case-by-case basis. It is not possible to propose 611 a default value for this UF, as it will be directly dependent on the dataset available. 612

613

Extrapolation from subchronic to chronic duration: the EFSA Scientific Committee recommends the 614 use of an UF of 2, considering the extent of investigations usually performed in 90-day studies. 615

Extrapolation from subacute to chronic duration: the Scientific Committee is not in a position to 616 propose default values, due to differences in the respective study designs. 617

618

Absence of a NOAEL: if the dataset allows for applying the BMD approach, there is no need for 619 applying any additional UF. In cases where the BMD approach cannot be applied, the LOAEL 620 approach will be used and an additional UF will be needed, the size of which should be determined on 621 a case-by-case basis. 622

623

5.3. Severity and nature of the observed effect 624

Severity is the degree to which an effect changes and impairs the functional capacity of an organ 625 system. Whilst the application of an additional UF because of the severity of an effect is not routinely 626 used, there are some examples where such a factor was considered necessary. 627


20

In EHC 240 (IPCS, 2009), the use of an extra UF for the severity of the effect is recommended for the 628 derivation of the ARfD by JMPR: if a toxicological effect is judged to be irreversible or particularly 629 severe, this should be a trigger to consider the finding in more detail before choosing an appropriate 630 UF. 631

According to the Guidelines for Drinking water Quality (WHO, 2008), an additional UF might be 632 applied when the end point is a foetal malformation, or when the endpoint determining the NOAEL 633 (or BMDL) is directly related to possible carcinogenicity. 634

In the field of pesticides, the following principle is described in the legislation, with regard to the 635 establishment of reference values (ADI, ARfD and AOEL): “When the critical effect is judged of 636 particular significance, such as developmental neurotoxic or immunotoxic effects, an increased margin 637 of safety shall be considered, and applied if necessary” (Regulation n°1107/2009). When an additional 638 UF has been considered necessary, a factor of up to 10 was applied. 639

640

For genotoxic and carcinogenic compounds which may be found in food, irrespective of their origin, 641 and where no health-based guidance values can be established, the Scientific Committee of EFSA 642 recommends the margin of exposure (MOE) approach (EFSA, 2005). The MOE is the ratio between a 643 reference point on the dose-response curve for the adverse effect and the human intake; as such, it 644 does not make implicit assumptions about a “safe” intake. In this opinion, the Scientific Committee 645 proposed that a MOE of 10,000 or higher, if based on the BMDL10

7 from an animal study, and taking 646 into account overall uncertainties in the interpretation, would be of low concern from a public health 647 point of view. However, such a judgement is ultimately a matter for the risk managers. Moreover a 648 MOE of that magnitude should not preclude the application of risk management measures to reduce 649 human exposure. Similarly, under the new European chemicals regulation (REACH), the ECHA 650 guidance (2010) recommends to use a Derived-Minimal-Effect-Level (DMEL) for non-threshold 651 substances, which is a MOE approach. 652

653

Conclusions: 654

The Scientific Committee considers that the need for an extra UF to allow for the severity of an effect 655 is exceptional, and therefore recommends considering its use on a case-by-case basis. 656

For genotoxic and carcinogenic compounds where no health-based guidance values can be established, 657 the Scientific Committee refers the reader to the margin of exposure approach described in its opinion 658 related to a harmonised approach for risk assessment of substances which are both genotoxic and 659 carcinogenic (EFSA, 2005). 660

661

5.4. Probabilistic approaches and combinations of uncertainty factors 662

One alternative to the use of deterministic uncertainty factors in traditional risk assessment is the use 663 of probabilistic distributions of the UFs. As lognormal distributions are thought to best describe 664 variability and uncertainty in UFs, these distributions have been derived based on NOAEL-ratios from 665 comprehensive toxicological databases. Different methodologies in the literature have provided 666 estimates of UFs for the inter-/intra-species extrapolation, for the exposure duration extrapolation, for 667 the use of a LOAEL, and for the combination of the different UFs. 668

7 The BMDL10 is the 95th percent lower confidence limit of the BMD of 10% extra risk.


21

In the standard procedure of deterministic risk assessment, the point estimates of various UFs are 669 multiplied to obtain an overall UF. Due to the possible interdependence of several UFs (e.g. time 670 extrapolation and interspecies variability), multiplication of the single UFs may lead to possibly overly 671 conservative estimates. This cumulation of worst case assumptions can be avoided by using 672 probability distributions of the various UFs. Under the assumption that the distributions of the UFs are 673 independent, their combination can be modelled, e.g. using Monte Carlo simulation, yielding a 674 probability distribution of the overall UF. This offers the possibility for a quantitative estimate of the 675 probability that an adverse effect will occur in a certain population at the estimated exposure level. 676 Moreover, the distribution of the overall UF can be probabilistically combined with the distribution of 677 the BMD, as also the effect parameter is uncertain and is best described by a lognormal distribution 678 (ECHA, 2010). 679

Several limitations of these probabilistic approaches for the UFs are highlighted by Vermeire (2001), 680 based on the fact that all distributions proposed are based on analyses of historical data, i.e. NOAEL 681 ratios: 682

1) The criteria used for constructing databases are not always transparent and NOAEL-ratios may have 683 been assessed without knowing the quality of the underlying data. 684

2) The uncertainty in the NOAEL as an estimate of the true No-Adverse-Effect-Level (NAEL) is 685 unknown. If ratios of NAELs would have been used, the distributions would have been less wide (i.e. 686 smaller geometric standard deviation). 687

3) Although the proposed default distributions are considered sufficiently founded to justify their 688 application in human risk assessment, further research on the basis of larger databases is still 689 considered necessary, especially with regard to the intraspecies distribution. 690

4) In the derivation of an interspecies UF from NOAEL-ratios, it is assumed that variability between 691 laboratory animals represents animal-human variability. 692

The advantage of the probabilistic risk assessment is that of more accurate risk estimates consistent 693 with the probabilistic nature of risk, whereas the disadvantages are those of being demanding in terms 694 of data collection/availability, calculation effort and experience of the risk assessor. Other factors 695 limiting the use of probabilistic techniques are the lack of guidance on the approach, including the 696 selection of models. In addition, there are difficulties in interpretation of the computed outcome and 697 the related risk communication. For these reasons, the probabilistic risk assessment is usually 698 undertaken only for substances of high concern and large data availability (ECHA, 2010). 699

The probabilistic approach was used by the EFSA Panel on Contaminants in the Food Chain 700 (CONTAM) when establishing the Tolerable Weekly Intake (TWI) for cadmium (EFSA, 2009). The 701 Panel modelled summary data from the literature relating urinary cadmium concentration to urinary 702 beta-2-microglobulin (B2M), a biomarker of kidney function, in order to derive a reference point from 703 the reported subgroup means. Because the individual data were not available, the CONTAM Panel 704 divided its reference point by a data-derived adjustment factor to allow for individual variability 705 within the dose groups. In contrast, JECFA concluded that it could not be assumed that urinary B2M 706 concentrations would vary as a function of urinary cadmium concentration within a sub-group. 707 Therefore the JECFA modelled the toxicodynamic variability by introducing a log-triangular 708 distribution function with a fixed range of variation by a factor between 1 and 3 below and above a 709 reference point identified from the same data set. Individual values were generated in a Monte-Carlo 710 simulation approach for both increased and reduced individual susceptibility resulting in a distribution 711 around the reference point. 712

713


22

Conclusions: 714

The Scientific Committee recommends that these probabilistic approaches and combination of UFs are 715 further investigated before harmonisation is proposed within EFSA. 716

717

6. Rounding of figures when deriving health-based guidance values 718

Communicating an estimated figure (e.g. an HBGV) with an inadequate number of significant figures 719 may convey a spurious idea of precision, masking therefore the assumptions made and the uncertainty 720 factors that were used to establish the HBGV. 721

When dealing with a measured value, the degree of precision is determined by the precision of the 722 analytical methodology. When reporting derived values, then the degree of precision should take into 723 account the precision of the components used in the derivation. As a general rule, rounding should 724 happen as late as possible in the assessment process, for example in establishing an ADI. 725

The Scientific Committee emphasizes that the following rule should be applied for rounding a value: 726 If the digit to the right of the last significant digit is less than 5, that last significant digit is not 727 changed. If the digit to the right of the last significant digit is 5 or greater, that last significant digit is 728 rounded up. A digit is defined as significant if it contributes to the precision of the value, which 729 excludes: 730

• Leading zeros where they serve merely to indicate the scale of the number (e.g. 0.006 has one 731 significant figure). 732

• Spurious digits introduced, for example, by calculations carried out to greater accuracy than 733 that of the original data, or measurements reported to a greater precision than the equipment 734 supports (e.g. 2.000). 735

736

The practical impact of rounding an health-based guidance value will vary depending on the numerical 737 closeness of the unrounded value to the rounded value, for example the impact of rounding 1.9 to 2 is 738 less than of rounding 1.4 to 1. The latter case could present difficulties for risk managers if, for 739 example, exposure was estimated to be 1.3 mg/kg b.w. per day. One approach to dealing with this 740 would be to round to a single significant figure if the impact of rounding is less than a certain 741 percentage, and to two significant figures if the impact exceeds that percentage. 742

The measurement of an adverse health effect differs from a chemical analysis in the sense that the 743 precision of the measurement of an effect (or the power to detect an effect) in an animal study is 744 determined by a number of factors, including the numbers of animals per dose group, variability in 745 dose to individual animals throughout the duration of a study (e.g. if the chemical is administered in 746 the diet to animals caged in groups), dose spacing (if a NOAEL approach is used rather than a 747 BMDL), as well as the measurement method used to detect the toxicological endpoint of interest. 748 Consequently, the overall precision is unlikely to be less than 10%. Table 8 illustrates rounding to one 749 or two significant figures such that the rounded figure does not vary by more than 10% from the 750 unrounded figure, and hence is likely to be within the range of experimental error. 751


23

Table 8: Impact of a 10% variation threshold as a rule for rounding 752

Unrounded figure

Rounded to one significant figure

% change Proposed rounding

% change

0.098 0.1 2.0 0.1 2.0 0.268 0.3 11.9 0.27 0.7 1.784 2 12.1 1.8 0.9 1.839 2 8.8 2 8.8 5.198 5 -3.8 5 -3.8 14.86 10 -32.7 15 0.9 26.24 30 14.3 26 -0.9 346.3 300 -13.4 350 1.1

753

Conclusions: 754




762

CONCLUSIONS AND RECOMMENDATIONS 763

The purpose of this guidance document is harmonisation of default values used by EFSA Scientific 764 Panels and Committee, and not standardisation: it is therefore always possible to deviate from the 765 proposed default values, provided that the rationale for such deviation is described. 766

Following the review of default values used by the EFSA Scientific Panels and Committee, and the 767 EFSA Units, and the most recent national information compiled in the EFSA Comprehensive 768 European Food Consumption Database, the Scientific Committee recommends the following default 769 values to be used in the absence of empirical data: 770

Body weight (See section 2) 771

• A body weight of 70 kg should be used as default for European adults. 772

• For dietary exposure assessment, a body weight of 12 kg should be used as default for 773 European infants and children. If deviation from the default value is required for the 774 assessment of specific age groups, the median values identified in table 3 should be used. 775

776


24

Total food and liquid intake (see section 3) 777

• The Scientific Committee was not in a position to propose a harmonised default value for 778 daily total solid food intake for adults. Such values should be considered according to the 779 relevant guidelines and legal requirements. The Scientific Committee however recommends 780 that the EFSA Comprehensive Database is regularly consulted to check the relevance of the 781 default values that are used. The Scientific Committee did not consider possible default values 782 for daily total solid food intake for children, as the EFSA Panels and Units did not report 783 using such default values. 784

• A 2 L default value for daily total liquid intake is recommended for European adults, 785 including the elderly. For children’s total liquid intake, the reader is referred to the NDA 786 opinion on dietary reference values for water where adequate intake levels are provided 787 according to their age. 788

789

Factors for converting chemical compound concentrations in feed or drinking water into daily 790 doses in experimental animal studies (see section 4) 791

Administration of test compounds in feed 792

• For chronic studies, a default factor of 0.05 for rats and 0.15 for mice, e.g. 1 mg/kg in feed is 793 equivalent to a dose of 0.05 and 0.15 mg/kg bw per day in rats and mice, respectively 794

• For subacute studies, a default factor of 0.12 for rats and 0.2 for mice 795

• For subchronic studies, a default factor of 0.09 for rats and 0.2 for mice 796

• Experiments start usually at week 5 to 7. If a subacute or a subchronic study starts at a later 797 age, e.g. after week 25, then the conversion factor for chronic studies should be applied. 798

799

Administration of test compounds in drinking water 800

• For chronic studies, a default factor of 0.05 for rats and 0.09 for mice, e.g. 1 mg/L in water is 801 equivalent to a dose of 0.05 and 0.09 mg/kg bw per day in rats and mice, respectively 802

• For subacute studies, a default factor of 0.12 for rats and 0.18 for mice 803

• For subchronic studies, a default factor of 0.09 for rats and 0.15 for mice 804

• Experiments start usually at week 5 to 7. If a subacute or a subchronic study starts at a later 805 age, e.g. after week 25, then the conversion factor for chronic studies should be applied. 806

807

Uncertainty factors (UFs) used in establishing health-based guidance values (See section 5) 808

Intra/inter-species extrapolation 809

• In the absence of chemical-specific data on kinetics and/or dynamics, the Scientific 810 Committee recomends using the overall default uncertainty factor of 100 (10 for inter-species 811 variability x 10 for intra-human variability). 812


25

• If available, chemical-specific data on kinetics and/or dynamics should be used. For the 813 remaining components, for which data are not available, the following default sub-factors 814 should be applied: 815

o for inter-species variability in toxicokinetics: 4.0 816

o for inter-species variability in toxicodynamics: 2.5 817

o for intra-human variability in toxicokinetics: 3.16 818

o for intra-human variability in toxicodynamics: 3.16 819

820

Deficiencies in the data available for the assessment 821

• In case of deficiencies in the data for applications, the EFSA Scientific Committee 822 recommends that, rather than applying an additional UF, the possibility/feasibility of getting 823 additional data improving the quality of the dataset is first considered. Therefore, the lack of 824 data should not directly imply the use of an extra UF. 825

• When additional data cannot be obtained or requested, the use of an additional UF to take 826 account of the deficiency of a database should be considered on a case-by-case basis. It is not 827 possible to propose a default value for this UF, as it will be directly dependent on the dataset 828 available. 829

830

Extrapolation for duration of exposure 831

• Extrapolation from subchronic to chronic duration: the EFSA Scientific Committee 832 recommends the use of an UF of 2, considering the extent of investigations usually performed 833 in 90-day studies. 834

• Extrapolation from subacute to chronic duration: the Scientific Committee is not in a position 835 to propose default values, due to differences in the respective study designs. 836

837

Accounting for the absence of a NOAEL 838

• In its previous opinion, the Scientific Committee recommended using the benchmark dose 839 (BMD) approach rather than the NOAEL or LOAEL for deriving the reference point. When 840 using the BMD approach, there is then no need for an additional UF. 841

• In cases where the BMD approach cannot be applied and there is no NOAEL for the critical 842 effect, the LOAEL can be used. In this case, an additional UF is needed, the size of which 843 should be determined on a case-by-case basis. 844

845

Severity and nature of the observed effect 846

• The Scientific Committee considers that the need for an extra UF to allow for the severity of 847 an effect is exceptional, and therefore recommends considering its use on a case-by-case basis. 848


26

• For genotoxic and carcinogenic compounds where no health-based guidance values can be 849 established, the Scientific Committee refers the reader to the margin of exposure approach. 850

851

Probabilistic approaches and combinations of uncertainty factors 852

The Scientific Committee considered the use of probabilistic distribution of UFs for EFSA’s risk 853 assessment, as an alternative for multiplying various UFs for deriving health-based guidance values, 854 which may end up in cumulating worst case assumptions. The Scientific Committee recommends that 855 these probabilistic approaches and combination of UFs are further investigated before harmonisation is 856 proposed within EFSA. 857

858

Rounding of figures when deriving health-based guidance values (see section 6) 859




867

REFERENCES 868

Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (COT), 2007. 869 Variability and Uncertainty in Toxicology of Chemicals in Food, Consumer Products and the 870 Environment. UK - March 2007. 871 (http://www.food.gov.uk/science/ouradvisors/toxicity/COTwg/wgvut/) 872

European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC), 2001. Technical Report 873 N°79. Exposure Factors Sourcebook for European Populations (with Focus on UK Data). Available 874 online: www.ecetoc.org 875

EFSA Panel on Contaminants in the Food Chain (CONTAM), 2009. Scientific Opinion on a request 876 from the European Commission on cadmium in food. The EFSA Journal (2009) 980, 1-3. 877 doi:10.2903/j.efsa.2009.980. Available online: www.efsa.europa.eu 878

EFSA Panel on Dietetic Products, Nutrition, and Allergies (NDA), 2010. Scientific opinion on dietary 879 reference values for water. The EFSA Journal 2010; 8(3):1459. [48 pp.]. 880 doi:10.2903/j.efsa.2010.1459. Available online: www.efsa.europa.eu 881

EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF), 2010. 882 Statement on the Safety Evaluation of Smoke Flavourings Primary Products: Interpretation of the 883 Margin of Safety. The EFSA Journal 2010; 8(1):1325. doi:10.2903/j.efsa.2009.1325. Available 884 online: www.efsa.europa.eu 885

EFSA Scientific Committee (SC), 2005. Scientific opinion on a request from EFSA related to a 886 harmonised approach for risk assessment of substances which are both genotoxic and carcinogenic. 887 The EFSA Journal (2005) 282, 1-31. doi:10.2903/j.efsa.2005.282. Available online: 888 www.efsa.europa.eu 889


27

EFSA Scientific Committee (SC), 2009. Guidance on a request from EFSA on the use of the 890 benchmark dose approach in risk assessment. The EFSA Journal (2009) 1150, 1-72. 891 doi:10.2903/j.efsa.2009.1150. Available online: www.efsa.europa.eu 892

European Chemicals Agency (ECHA), 2010. Guidance on information requirements and chemical 893 safety assessment. Chapter R.8: Characterisation of dose[concentration]-response for human health. 894 Guidance for implementation of REACH (http://echa.europa.eu/). 895

European Commission, 2001. Guidelines of the Scientific Committee on Food for the presentation of 896 an application for safety assessment of a substance to be used in food contact materials prior to its 897 authorisation. http://ec.europa.eu/food/fs/sc/scf/out82_en.pdf 898

Falk-Filipsson, A., Hanberg, A., Victorin, K., Warholm, M., Wallén, M., 2007. Assessment factors – 899 Applications in health risk assessment of chemicals. Environmental Research, 104: 108-127. 900

ICRP (International Commission on Radiological Protection), 1974. Report of the Task Group on 901 Reference Man. ICRP Publication No. 23. Oxford, Pergamon Press. 902

Interdepartmental Group on Health Risks from Chemicals (IGHRC), 2003. Uncertainty factors: Their 903 use in human health risk assessment by UK government. Institute for Environment and Health – 904 IEH, UK (http://www.le.ac.uk/ieh/). 905

IPCS (International Programme on Chemical Safety), 1987. Environmental Health Criteria 70, Food 906 additives and contaminants in food, principles for the safety assessment of food additives and 907 contaminants in food. Geneva, Switzerland, World Health Organization, International Programme 908 on Chemical Safety. 909

IPCS (International Programme on Chemical Safety), 1994. Environmental Health Criteria 170: 910 Assessing human health risks to chemicals: derivation of guidance values for health-based 911 exposure limits. Geneva, Switzerland, World Health Organization, International Programme on 912 Chemical Safety. 913

IPCS (International Programme on Chemical Safety), 1999. Environmental Health Criteria 210: 914 Principles for the assessment of risks to human health from exposure to chemicals. Geneva, 915 Switzerland, World Health Organization, International Programme on Chemical Safety. 916

IPCS (International Programme on Chemical Safety), 2009. Environmental Health Criteria 240, 917 Principles and Methods for the Risk Assessment of Chemicals in Food. Geneva, Switzerland, Food 918 and Agriculture Organization / World Health Organization, International Programme on Chemical 919 Safety. 920

Lehman, A.J., 1954. Approximate relation of parts per million in the diet to mg/kg D.W./day. 921 Association of Food and Drug Officials Quarterly Bulletin, 18:66 . 922

Luijckx N.B.L., Rao G.N., McConnell E.E., Würtzen G. and Kroes R., 1994. The intake of chemicals 923 related to age in long-term toxicity studies – considerations for risk assessment. Regulatory 924 Toxicology and Pharmacology, 20: 96-104. 925

NTP: U.S. National Toxicology Program Long-Term Study Reports & Abstracts 926 (http://ntp.niehs.nih.gov/?objectid=D16D6C59-F1F6-975E-7D23D1519B8CD7A5) latest access 927 January 2011. 928

Vermeire, T., Pieters, M., Rennen, M., Bos, P., 2001. Probabilistic assessment factors for human 929 health risk assessment. RIVM Report 601516005, TNO Report V3489. 930

World Health Organisation (WHO), 2008. Guidelines for Drinking-water Quality – Third edition 931 incorporating the first and second addenda. 932 (http://www.who.int/water_sanitation_health/dwq/gdwq3rev/en/) 933

World Health Organization (WHO)/International Programme on Chemical Safety (IPCS), 2005. 934 Chemical-Specific Adjustment Factors for Interspecies Differences and Human Variability: 935


28

Guidance document for Use of Data in Dose/Concentration –Response Assessment. Harmonisation 936 Project document N°2. 937

Zarn, J.A., Hänggi, E., Kuchen, A., Schlatter, J.R., 2010. The significance of the subchronic toxicity in 938 the dietary risk assessment of pesticides. Regulatory Toxicology and Pharmacology, 58: 72-78. 939

940

PUBLIC CONSULTATION DRAFT guidance on Default assumptions

29

APPENDIX 941 Default Values proposed for use by EFSA Scientific Panels and Committee, and EFSA Units 942

943 Issue for harmonisation Default value proposed Remark

Body weight (Kg) Adults: 70 Infants & children: 12 Total solid food intake No default value Total liquid intake (L) Adults: 2 Children: no default values For children,

See http://www.efsa.europa.eu/en/efsajournal/doc/1459.pdf Converting test compound concentrations in feed (mg/kg), into daily dose (mg/kg b.w. per day)




Converting test compound concentrations in drinking water (mg/l), into daily dose (mg/kg b.w. per day)




UF for inter-species extrapolation


Variability in toxicokinetics: 4.0 Variability in toxicodynamics: 2.5

UF for intra-human extrapolation


variability in toxicokinetics: 3.16 variability in toxicodynamics: 3.16

UF for Deficiencies in the data available for the assessment

No default UF

Consider the possibility/feasability of getting additional data first. If not feasible, use additional UF (value determined on a case-by-case basis).

UF for for duration of exposure extrapolation

Subchronic to chronic: 2 Subacute to chronic: no default UF

UF to account for the absence of a NOAEL

No default UF Use the BMD approach. If not possible, consider use of the LOAEL with an additional UF (value determined on a case-by-case basis)

UF to account for the severity and nature of the effect

No default UF Genotoxic and carcinogenic compounds: Use the Margin of Exposure approach

Usually not needed. If exceptionally considered necessary, UF value determined on a case-by-case basis.

UF: Uncertainty Factor 944


30

945

ABBREVIATIONS 946

ADI Acceptable Daily Intake ANS EFSA Panel on Food Additives and Nutrient Sources Added to Food AOEL Acceptable Operator Exposure Level ARfD Acute Reference Dose BMD Benchmark Dose BMDL Lower confidence limit of the BMD CEF EFSA Panel on Food Contact Materials, Enzymes, Flavourings and

Processing Aids CONTAM EFSA Panel on Contaminants in the Food Chain CSAF Chemical-Specific Adjustment Factors DCM EFSA Dietary and Chemical Monitoring Unit (former DATEX Unit) DMEL Derived-Minimal-Effect-Level ECHA European Chemicals Agency EHC Environmental Health Criteria FAO Food and Agriculture Organization FEEDAP EFSA Panel on Additives and Products or Substances used in Animal Feed HBGV Health-Based Guidance Value ICRP International Commission on Radiological Protection IPCS International Programme on Chemical Safety JECFA Joint FAO/WHO Expert Committee on Food Additives JMPR Joint FAO/WHO Meeting on Pesticide Residues LOAEL Lowest-Observed-Adverse-Effect-Level MOE Margin of Exposure NAEL No-Adverse-Effect-Level NDA EFSA Panel on Dietetic Products, Nutrition and Allergies NOAEL No-Observed-Adverse-Effect-Level NTP U.S. National Toxicology Program PPP Plant Protection Products PPR EFSA Panel on Plant Protection Products and their Residues PRAS EFSA Pesticides Unit REACH Registration, Evaluation, Authorisation and Restriction of Chemicals TDI Tolerable Daily Intake TTC Threshold of Toxicological Concern TWI Tolerable Weekly Intake UF Uncertainty Factor WHO World Health Organisation 947

Documents

Guidance on Default assumptions used by the EFSA ...PUBLIC CONSULTATION DRAFT guidance on Default assumptions 3 52 Default Values proposed for use by EFSA Scientific Panels and Committee,